Relations of interannual differences in stream litter breakdown with discharge: bioassessment implications

Ecosystem-level processes are increasingly used by researchers and managers as indicators of ecological integrity for bioassessment, particularly in streams. However, processes like litter breakdown integrate multiple mechanistic pathways, which can vary differentially even under natural, unimpacted conditions. In particular, weather-driven hydrologic variations may strongly influence invertebrate shredder feeding and physical abrasion, inducing high natural variability of litter breakdown rates, which may obscure the effects of anthropogenic disturbances. Yet, such variability is rarely assessed to refine benchmarks of ecological status. Here, we quantified how interannual hydrologic differences contributed to the spatio-temporal variability of litter breakdown rate and its components (fragmentation, lambda(F); and dissolution and microbial decomposition, lambda(m)), in low-order unimpacted, perennial streams across three climatically similar regions in temperate Canada. We measured litter breakdown rates in coarse (5 or 10mm; k(c))- and fine-mesh (0.5mm; k(f)) leaf bags during fall for 2-4 yr and used variance partitioning to disentangle the variation of k(c), lambda(F), and lambda(m), as explained by hydrologic indices (during and prior to leaf bag incubation), decomposer-related variables, and water chemistry. Contrary to our hypotheses, interannual hydrologic variability was unrelated to lambda(F), and poorly predicted lambda(m) and k(c) within regions. Within-region spatial (i.e., across sites in a year) and temporal (across years at a site) differences in k(c) approximated or exceeded the range of natural variability suggested to characterize reference conditions by a popular bioassessment framework. Accordingly, we recommend site- and region-specific modifications of benchmarks for reference conditions that account for interannual variability, while cautioning about their potential non-stationarity under climate change. Composite parameters such as k(c)/k(f) and lambda(F)/lambda(m) were substantially more variable across sites, and hence are not robust assessment parameters. As the range of natural variability of litter breakdown revealed in this study can overlap with the average impacts of certain anthropogenic disturbances (e.g., nutrient enrichment) on this parameter reported by previous research, we emphasize the need to include other structural and functional indicators to ensure comprehensive stream bioassessments.